Data of red deer and moose space use in a confined nature area

This data set and R script were used to analyze (1) step length and (2) habitat selection by red deer and moose in a fenced nature area in Denmark.To analyze variation in hourly step length (log-transformed, response variable), we used generalized additive models with a Gamma distribution and log-link of the R package ‘mgcv’, separately for red deer and moose and for the 4 seasons. We included the hour of the day and the number of daily visitors as smooth terms, daily mean temperature and sex as parametric effects, and individual ID as random effect to account for varying numbers of GPS locations and individual variation. Moreover, we included the interaction between daily visitor numbers and time of day to quantify if red deer and moose adjust diel movement conditional on daily visitor numbers. To analyze habitat selection (1 = actual step versus 0 = random step), we used generalized mixed-effects models of R package ‘lme4’ with a binomial family and logit link, including step ID nested within individual ID as a random effect to make sure the model compared each actual step to the generated random steps that belonged to it. We conducted separate analyses for the two species, two behavioral states (inactive versus active) and the 4 seasons to avoid fourth-order interactions. For each of the 16 analyses, we compared 5 candidate models: (1) a basic model including sex, average daily temperature, land cover, the distance to roads, trails, the boundary fence, and the interaction of sex with land cover and of temperature with land cover; (2) a model quantifying if habitat selection was conditional on the time of the day as proxy for visitor intensity, by including visitor intensity (see categorization below) as two-way interaction terms with the above-mentioned habitat variables; (3) a model quantifying if habitat selection was conditional on daily visitor numbers (independent of time of day), including the number of daily visitors as two-way interaction terms with the habitat variables; (4) a combined model including both two-way interactions of the habitat variables with daily visitor numbers and visitor intensity (based on time of the day); and (5) a model quantifying if habitat selection was conditional on daily visitor numbers as well as time of day, including the three-way interaction of the daily visitor numbers, visitor intensity, and the habitat variables. We categorized visitor intensity based on the time of the day: (1) midday, from 10:00 to 15:00, representing high visitor numbers; (2) mornings and evenings, from 6:00 to 9:00 and 16:00 to 20:00, representing low to intermediate visitor numbers; and (3) night, from 21:00 to 5:00, no visitors present.